The talk starts with a short introduction on work I presented so far during my first year at the IRTG. I then describe how this work was recently extended to speech laughs. Finally, I present some thoughts and ideas of what could be other enhancements of the application of articulatory synthesis to the periphery of the core synthesis task of words and sentences.

Speech laughs occur much more frequently in everyday communication than we would think. They are often very short or subtle variations of the regular flow of our speech. Although they are similar to laughs, the difference is that the "linguistic" utterance is running in parallel to the "emotive" one. Among the characteristic features are for example inserted or increased friction noises, vocal tremors, smiled phonemes, or syllabic pulsation.
I analyzed the acoustic properties of a range of speech laughs and built synthetic variants of one speech laugh sample. Participants of a perception test assessed the degree of amusement and the naturalness of isolated test stimuli. Results indicate that syllabic pulsation influences the rating of amusement whereas smiled vowel quality does not. Low naturalness ratings suggest further testing including the context of a speech laugh.

Another trial application of articulatory speech synthesis could be to emulate regional accents of German to find out whether systematic variation exists across regions and how it could be described in an economic way. I am going to give a brief introduction on how this topic is related to previous work and present some aspects a future study could cover.